US 8072541 B2 Abstract A signal processing apparatus converts interlaced 4:2:2-format video signals including a luminance component and two chroma components into interlaced 4:2:0 format video signals. A downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25. The downsampling low-pass filter is designed together with an associated upsampling low-pass filter such that the first downsampling low-pass filter and the upsampling low-pass filter substantially satisfy perfect reconstruction filter bank condition, and such that a sum of the group delay of the first downsampling low-pass filter and a group delay of a normalized filter obtained by making the sum of all the coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0.
Claims(12) 1. A signal processing apparatus for converting interlaced 4:2:2-format video signals including a luminance component and two chroma components into interlaced 4:2:0 format video signals, comprising:
a first downsampling unit configured to apply vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:2-format video signals by use of a first downsampling low-pass filter and to downsample by a ratio of 2:1 to produce a first chroma field of the interlaced 4:2:0-format video signals; and
a second downsampling unit configured to apply vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:2-format video signals by use of a second downsampling low-pass filter obtained by reversing an order of coefficients of the first downsampling low-pass filter and to downsample by a ratio of 2:1 to produce a second chroma field of the interlaced 4:2:0-format video signals,
wherein the first downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25,
wherein the first downsampling low-pass filter is designed together with an associated upsampling low-pass filter such that the first downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, and such that a sum of the group delay of the first downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range.
2. The signal processing apparatus as claimed in
3. A signal processing apparatus for converting interlaced 4:2:0-format video signals including a luminance component and two chroma components into interlaced 4:2:2 format video signals, comprising:
a first upsampling unit configured to apply vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:0-format video signals by use of a first upsampling low-pass filter and to upsample by a ratio of 1:2 to produce a first chroma field of the interlaced 4:2:2-format video signals; and
a second upsampling unit configured to apply vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:0-format video signals by use of a second upsampling low-pass filter obtained by reversing an order of coefficients of the first upsampling low-pass filter and to upsample by a ratio of 1:2 to produce a second chroma field of the interlaced 4:2:2-format video signals,
wherein the first upsampling low-pass filter is designed together with an associated downsampling low-pass filter such that the first upsampling low-pass filter and the downsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, such that the downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25, and such that a sum of the group delay of the downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the first upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range.
4. The signal processing apparatus as claimed in
5. A signal processing apparatus for converting interlaced 4:2:2-format video signals including a luminance component and two chroma components into interlaced 4:2:0 format video signals, comprising:
a filtering and downsampling unit; and
a coefficient storage configured to store coefficients of a first downsampling low-pass filter,
wherein the filtering and downsampling unit is configured to apply filtering to pixels in a first chroma field of the interlaced 4:2:2-format video signals by use of the coefficients of the first downsampling low-pass filter to produce a first chroma field of the interlaced 4:2:0-format video signals;
wherein the filtering and downsampling unit is configured to apply filtering to pixels in a second chroma field of the interlaced 4:2:2-format video signals by use of second downsampling low-pass filter coefficients obtained by reversing an order of the coefficients of the first downsampling low-pass filter to produce a second chroma field of the interlaced 4:2:0-format video signals,
wherein the first downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25,
wherein the first downsampling low-pass filter is designed together with an associated upsampling low-pass filter such that the first downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, and such that a sum of the group delay of the first downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range.
6. The signal processing apparatus as claimed in
7. A signal processing apparatus for converting interlaced 4:2:0-format video signals including a luminance component and two chroma components into interlaced 4:2:2 format video signals, comprising:
a filtering and upsampling unit; and
a coefficient storage configured to store coefficients of a first upsampling low-pass filter,
wherein the filtering and upsampling unit is configured to apply filtering to pixels in a first chroma field of the interlaced 4:2:0-format video signals by use of the coefficients of the first upsampling low-pass filter to produce a first chroma field of the interlaced 4:2:2-format video signals;
wherein the filtering and upsampling unit is configured to apply filtering to pixels in a second chroma field of the interlaced 4:2:0-format video signals by use of second upsampling low-pass filter coefficients obtained by reversing an order of the coefficients of the first upsampling low-pass filter to produce a second chroma field of the interlaced 4:2:2-format video signals,
wherein the first upsampling low-pass filter is designed together with an associated downsampling low-pass filter such that the first upsampling low-pass filter and the downsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, such that the downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25, and such that a sum of the group delay of the downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the first upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range.
8. The signal processing apparatus as claimed in
9. A signal processing method of converting interlaced 4:2:2-format video signals including a luminance component and two chroma components into interlaced 4:2:0 format video signals, comprising:
applying vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:2-format video signals by use of a first downsampling low-pass filter to perform downsampling to produce a first chroma field of the interlaced 4:2:0-format video signals; and
applying vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:2-format video signals by use of a second downsampling low-pass filter obtained by reversing an order of coefficients of the first downsampling low-pass filter to perform downsampling to produce a second chroma field of the interlaced 4:2:0-format video signals,
wherein the first downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25,
wherein the first downsampling low-pass filter is designed together with an associated upsampling low-pass filter such that the first downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, and such that a sum of the group delay of the first downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range.
10. The signal processing method as claimed in
11. A signal processing method of converting interlaced 4:2:0-format video signals including a luminance component and two chroma components into interlaced 4:2:2 format video signals, comprising:
applying vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:0-format video signals by use of a first upsampling low-pass filter to perform upsampling to produce a first chroma field of the interlaced 4:2:2-format video signals; and
applying vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:0-format video signals by use of a second upsampling low-pass filter obtained by reversing an order of coefficients of the first upsampling low-pass filter to perform upsampling to produce a second chroma field of the interlaced 4:2:2-format video signals,
wherein the first upsampling low-pass filter is designed together with an associated downsampling low-pass filter such that the first upsampling low-pass filter and the downsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, such that the downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25, and such that a sum of the group delay of the downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the first upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range.
12. The signal processing method as claimed in
Description The present application is based upon and claims the benefit of priority from the prior Japanese Patent Applications No. 2008-061174 filed on Mar. 11, 2008 and 2008-183456 filed on Jul. 15, 2008, with the Japanese Patent Office, the entire contents of which are incorporated herein by reference. The disclosures herein relate to a video signal processing apparatus for converting a video signal in one format to a video signal in another format. Video signals for moving pictures include a luminance signal and two chroma signals Cb and Cr. There are several formats for varying chroma signal resolutions. The 4:2:2 format is typically used for a digital interface in video equipment. An example of this video format includes the SMPTE 292M standard (i.e., HD-SDI) defined by the SMPTE (Society of Motion Picture and Television Engineers) For the purpose of video encoding in consumer products, on the other hand, the 4:2:0 format is typically used. - [Patent Document 1] Japanese Patent Application Publication No. 2000-92512
- [Patent Document 2] Japanese Patent No. 3292486
- [Non-Patent Document 1] “Problem and Solution of Current TV/HDTV Compatible Encoding Scheme,” The Institute of Electronics, Information and Communication Engineers, Proceedings of Spring Conference, D-334, 1992
According to one embodiment, a signal processing apparatus for converting interlaced 4:2:2-format video signals including a luminance component and two chroma components into interlaced 4:2:0 format video signals includes a first downsampling unit configured to apply vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:2-format video signals by use of a first downsampling low-pass filter and to downsample by a ratio of 2:1 to produce a first chroma field of the interlaced 4:2:0-format video signals, and a second downsampling unit configured to apply vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:2-format video signals by use of a second downsampling low-pass filter obtained by reversing an order of coefficients of the first downsampling low-pass filter and to downsample by a ratio of 2:1 to produce a second chroma field of the interlaced 4:2:0-format video signals, wherein the first downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25, wherein the first downsampling low-pass filter is designed together with an associated upsampling low-pass filter such that the first downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, and such that a sum of the group delay of the first downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. According to one embodiment, a signal processing apparatus for converting interlaced 4:2:0-format video signals including a luminance component and two chroma components into interlaced 4:2:2 format video signals includes a first upsampling unit configured to apply vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:0-format video signals by use of a first upsampling low-pass filter and to upsample by a ratio of 1:2 to produce a first chroma field of the interlaced 4:2:2-format video signals, and a second upsampling unit configured to apply vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:0-format video signals by use of a second upsampling low-pass filter obtained by reversing an order of coefficients of the first upsampling low-pass filter and to upsample by a ratio of 1:2 to produce a second chroma field of the interlaced 4:2:2-format video signals, wherein the first upsampling low-pass filter is designed together with an associated downsampling low-pass filter such that the first upsampling low-pass filter and the downsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, such that the downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25, and such that a sum of the group delay of the downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the first upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. According to one embodiment, a signal processing method of converting interlaced 4:2:2-format video signals including a luminance component and two chroma components into interlaced 4:2:0 format video signals includes applying vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:2-format video signals by use of a first downsampling low-pass filter to perform downsampling to produce a first chroma field of the interlaced 4:2:0-format video signals, and applying vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:2-format video signals by use of a second downsampling low-pass filter obtained by reversing an order of coefficients of the first downsampling low-pass filter to perform downsampling to produce a second chroma field of the interlaced 4:2:0-format video signals, wherein the first downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25, wherein the first downsampling low-pass filter is designed together with an associated upsampling low-pass filter such that the first downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, and such that a sum of the group delay of the first downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. According to one embodiment, a signal processing method of converting interlaced 4:2:0-format video signals including a luminance component and two chroma components into interlaced 4:2:2 format video signals includes applying vertical-direction filtering to pixels in a first chroma field of the interlaced 4:2:0-format video signals by use of a first upsampling low-pass filter to perform upsampling to produce a first chroma field of the interlaced 4:2:2-format video signals, and applying vertical-direction filtering to pixels in a second chroma field of the interlaced 4:2:0-format video signals by use of a second upsampling low-pass filter obtained by reversing an order of coefficients of the first upsampling low-pass filter to perform upsampling to produce a second chroma field of the interlaced 4:2:2-format video signals, wherein the first upsampling low-pass filter is designed together with an associated downsampling low-pass filter such that the first upsampling low-pass filter and the downsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range, such that the downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25, and such that a sum of the group delay of the downsampling low-pass filter and a group delay of a normalized filter obtained by making a sum of all coefficients of the first upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. The object and advantages of the embodiment will be realized and attained by means of the elements and combinations particularly pointed out in the claims. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not restrictive of the invention, as claimed. Broadcast equipment relays video signals successively from one location to another location. In such a case, the video signals may be encoded and decoded repeatedly because of the editing of tickers or the use of a switching apparatus for switching video signals at each location. To this end, an inexpensive 4:2:0-format-compatible encoder/decoder may be used. A 4:2:2-format video input With the use of filters satisfying the above-noted conditions, high-frequency components are inevitably degraded in the first-stage 4:2:2-to-4:2:0 conversion, but blurring will not spread more than the blurring of the output signals of the first stage even if a 4:2:0-to-4:2:2 conversion and a 4:2:2-to-4:2:0 conversion are repeatedly performed after the first stage. Upsampling of pixels by a ratio of 1:2 using the upsampling low-pass filter P(z) is performed by first inserting zero after each pixel, thereby converting an N-pixel reduced image into a 2N-pixel image. The upsampling low-pass filter is then applied to this 2N-pixel image to produce a 2N-pixel upsampled image. The above-described method is equivalent to the following process. An interpolation filter Pe(z) having the even-numbered coefficients of the upsampling low-pass filter P(z) and an interpolation filter Po(z) having the odd-numbered coefficients of the upsampling low-pass filter P(z) are constructed as follows.
In the following, a description will be given of what sort of relationships may preferably be satisfied by chroma pixel positions between before and after conversion with respect to a 4:2:2-to-4:2:0 format conversion and a 4:2:0-to-4:2:2 format conversion. As a background explanation, a description will first be given with regard to how large a positional displacement occurs for a pixel after filtering relative to an original pixel position. In order to represent a phase shift of the original signal, group delay characteristics will be used. The frequency characteristics of filter F are represented as follows.
Multiplication of all the terms of the z transform of a filter by z A simple way to derive a group delay with respect to ω=0 is to use the following formula wherein filter coefficients are denoted as f(n), and the sum of the coefficients is assumed to be 1.
Patent Document 1 discloses two sets of examples of A(z) and P(z) that are expressed as following formulas (9a) and (9b) as well as formulas (10a) and (10b).
In the following, a description will be given of why the filters disclosed in Patent Document 1 are not superior in terms of display quality by focusing attention on expressions (9a) and (9b). Downsampling low-pass filter A(z) defined by expression (9a) has a group delay of 0 with respect to ω=0 according to formula (7). As for upsampling low-pass filter P(z), Pe(z) and Po(z) are obtained as follows by using expression (9b) in formulas (5a) and (5b).
In the following, a description will be given of how a pixel is shifted after filtering by referring to In the following, a description will be given of a method for evaluating the pixel displacement occurring between before and after downsampling and upsampling when the filters defined by expressions (10a) and (10b) are combined with a conventional filter having preferable characteristics. This evaluation method is based on a point of view different from that of the above-described method. In order to examine the characteristics of expression (10b), the characteristics of a signal obtained by downsampling and upsampling will be examined in the following. To this end, the signal defined in expression (3) obtained by downsampling followed by upsampling is expanded.
In video equipment, no positional displacement of pixels between before and after downsampling and upsampling may be tolerable. Accordingly, the group delay of the filter defined by expression (14) with respect to ω=0 may preferably be 0 (or equal to an integer number). It is known to those skilled in the art that filter F(z)G(z) obtained by combining filter F(z) and filter G(z) has a group delay with respect to ω=0 that is a sum of the group delay of filter F(z) and the group delay of filter G(z) with respect to ω=0. For video equipment, thus, the sum of the group delays of ½P(z) and A(z) with respect to ω=0 may preferably be an integer number. An image downsampled by a preferable filter may be upsampled by a filter defined by expression (10b). In such a case, the group delay of the preferable filter with respect to ω=0 is ¼ as previously described. The group delay of the filter defined by expression (10b) with respect to ω=0 is ½. According to expression (14), the group delay occurring between before and after downsampling and upsampling is ¾, which indicates that pixels are shifted by as much as ¾ pixel through such processing. The use of such a filter creates color misrepresentation relative to the original image, thereby generating visually discernible artifacts that cannot be ignored. In order to avoid such artifacts, Patent Document 1 discloses using a cascade-connection-purpose filter and a “normal” filter having preferable group delay characteristics as previously described. As has been described heretofore, no filter has been conventionally known that satisfies both the first condition that image degradation does not worsen through cascade connections and the second condition that group delay characteristics equal to those of conventionally used display-purpose or size-reduction-purpose filters are provided. The absence of such a filter necessitates the selective use of either a cascade-connection purpose filter or a display/reduction-purpose filter. Further, all the perfect reconstruction filters that are conventionally known have the group delay equal to 0 or 0.5. Because of this, when a perfect reconstruction filter is used to downsample in interlaced scanning, the use of the same filter bank for the top and bottom fields to perform a 2:1 division into low-frequency signal L(z) and high-frequency signal H(z) causes the low-frequency signal to have an imperfect arrangement of interlaced scanning line positions. Namely, the pixel distance between the top field and the bottom field as well as the pixel distance between the bottom field and the top field are both unequal. This is described in detail in Non-Patent Document 1. In order to obviate this problem, Patent Document 2 discloses a method of equalizing the distance between the top field and the bottom field after downsampling. This method employs a filter (with even-number taps) satisfying the perfect reconstruction filter bank condition and having a group delay of 0.5 to downsample the top field and a filter (with odd-number taps) satisfying the perfect reconstruction filter bank condition and having a group delay of 0 to downsample the bottom field. However, the group delay characteristics of a desirable filter for converting 4:2:2 chroma into 4:2:0 chroma are equal to 0.25 for the top field and 0.75 (or −0.25) for the bottom field as illustrated in In the disclosures herein, a signal processing apparatus may advantageously prevent chroma pixels from having positional displacements when an image signal obtained by downsampling an interlaced chroma signal from the 4:2:2 format to the 4:2:0 format by use of a perfect reconstruction filter is displayed by use of a normal filter having preferable group delay characteristics. The applicants of the present application have found that a downsampling low-pass filter A(z) for converting a 4:2:2-format chroma signal into a 4:2:0-format chroma signal and a upsampling low-pass filter P(z) for converting the 4:2:0-format chroma signal into a 4:2:2-format chroma signal can be designed for the top field to simultaneously satisfy the first through fourth conditions defined in the following or simultaneously satisfy the first through third conditions. First Condition: the downsampling low-pass filter A(z) and the upsampling low-pass filter P(z) satisfy equations (1a), (1b), and either (1c) or (1e). This means that the perfect reconstruction filter bank condition are satisfied. This condition ensures that degradation does not worsen when upsampling and downsampling are serially cascaded.
In embodiments that follows, a description will be given of specific examples of a downsampling low-pass filter A(z) for converting a 4:2:2-format chroma signal into a 4:2:0-format chroma signal and an upsampling low-pass filter P(z) for converting the 4:2:0-format chroma signal into a 4:2:2-format chroma signal for the top field that simultaneously satisfy the first through fourth conditions defined above or simultaneously satisfy the first through third conditions. A filter satisfying the above-defined second condition may be modified by swapping higher-order coefficients with lower-order coefficients. That is, a filter satisfying the second condition and having coefficients at(n) (n=0, . . . , N−1) may be modified into a filter having coefficients a The use of such filters achieves the same group delay characteristics as the conventionally preferable characteristics in the conversion from the 4:2:2 format into the 4:2:0 format, so that the mixed use of encoding/decoding apparatuses having conventionally preferable filter characteristics and encoding/decoding apparatuses having filters of the disclosed embodiments does not create undesirable pixel positional displacements. Further, an increase in chroma degradation can be avoided even when the 4:2:2-to-4:2:0 conversion and the 4:2:0-to-4:2:2 conversion are repeated in cascade connections. According to at least one embodiment, no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. The applicants of the present application have successfully designed an 8-tap downsampling low-pass filter A(z) for converting a 4:2:2-format chroma signal into a 4:2:0-format chroma signal and an 8-tap upsampling low-pass filter P(z) for converting the 4:2:0-format chroma signal into a 4:2:2-format chroma signal for the top field that simultaneously satisfy the first through third conditions previously defined as well as equation (15) defined in the following, which is a general requirement for low-pass filters that the amplitudes of A(z) and P(z) be both equal to zero at frequency ω=π.
Coefficients p(k) of the 8-tap low-frequency synthesis filter satisfying equation (16) and the perfect reconstruction filter bank condition are given as follows.
P(z) is represented as follows by use of the coefficients given in coefficient list (17).
The 8-tap low-frequency analysis filter A(z) is represented as A(z)=Σa(k)·z The group delay of this low-frequency analysis filter A(z) is 0.223930156 at ω=0. This value is sufficiently close to an ideal value of 0.25 for practical purposes. The 8-tap low-frequency synthesis filter P(z) is represented as P(z)=Σp(k)·z Based on the above-described filter, coefficients pe(k) of the filter Pe(z) and coefficients po(k) of the filter Po(z) are given as follows.
A description will be given of a first embodiment by referring to A first downsampling unit A second downsampling unit The first downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25. The first downsampling low-pass filter is configured to allow at least one upsampling low-pass filter to exist such that the first downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range. The sum of the group delay of the first downsampling low-pass filter and the group delay of a normalized filter obtained by making the sum of all the coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. In this manner, the present embodiment employs a downsampling filter that satisfies the first through third conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. In A description will be given of a second embodiment by referring to A first upsampling unit A second upsampling unit The first upsampling low-pass filter is configured to allow at least one downsampling low-pass filter to exist such that the first upsampling low-pass filter and the downsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range. The downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25. The sum of the group delay of the downsampling low-pass filter and the group delay of a normalized filter obtained by making the sum of all the coefficients of the first upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. In this manner, the present embodiment employs an upsampling filter that satisfies the first through third conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. A description will be given of a third embodiment by referring to The 8 coefficients of the 8-tap top-field downsampling low-pass filter The coefficients of the downsampling low-pass filter The top-field downsampling low-pass filter is configured to allow at least one upsampling low-pass filter having the coefficients given in list (17) to exist such that the top-field downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range. The group delay of a filter obtained by normalizing the upsampling low-pass filter is 0.759511379 at frequency ω=0, and the group delay of the downsampling low-pass filter is 0.240488621 at frequency ω=0, so that the sum of these group delays is 1.00000000, which is substantially equal to an integer number. In this manner, the present embodiment employs a downsampling filter that satisfies the first through third conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. The present embodiment is directed to an example in which downsampling is performed after corresponding filtering. This is not a limiting example. Another configuration such as one in which filtering is selectively applied only to pixels to be output may alternatively be employed. A description will be given of a fourth embodiment by referring to The 8 tap coefficients of the top-field upsampling low-pass filter The coefficients of the bottom-field upsampling low-pass filter The downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is 0.240488621, which is substantially equal to 0.25. The group delay of the downsampling low-pass filter is 0.240488621 at frequency ω=0, and the group delay of a normalized filter obtained by making the sum of all the coefficients of the top-field upsampling low-pass filter equal to 1 is 0.759511379 at frequency ω=0, so that the sum of these two group delays is 1.00000000, which is substantially equal to an integer number. In this manner, the present embodiment employs an upsampling filter that satisfies the first through third conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. A description will be given of a fifth embodiment by referring to A downsampler The filter circuit The downsampler The first downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25. The first downsampling low-pass filter is configured to allow at least one upsampling low-pass filter to exist such that the first downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range. The sum of the group delay of the first downsampling low-pass filter and the group delay of a normalized filter obtained by making the sum of all coefficients of the upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. In this manner, the present embodiment employs a downsampling filter that satisfies the first through third conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. A description will be given of a sixth embodiment by referring to A filter circuit The upsampler The filter circuit The first upsampling low-pass filter is configured to allow at least one downsampling low-pass filter to exist such that the first upsampling low-pass filter and the downsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range. The downsampling low-pass filter has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25. The sum of the group delay of the downsampling low-pass filter and the group delay of a normalized filter obtained by making the sum of all coefficients of the first upsampling low-pass filter equal to 1 is substantially equal to an integer number at frequency ω=0 within a predetermined error tolerance range. In this manner, the present embodiment employs an upsampling filter that satisfies the first through third conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. A description will be given of a seventh embodiment by referring to The coefficients The filter circuit The coefficients The top-field downsampling low-pass filter having the coefficients given in list (18) has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is 0.223930156, which is 1 substantially equal to 0.25 for practical purposes. The top-field downsampling low-pass filter is configured to allow at least one upsampling low-pass filter having the coefficients given in list (19) to exist such that the top-field downsampling low-pass filter and the upsampling low-pass filter satisfy perfect reconstruction filter bank condition within a predetermined error tolerance range. The group delay of a filter obtained by normalizing the upsampling low-pass filter is 0.776069844 at frequency ω=0, and the group delay of the top-field downsampling low-pass filter is 0.223930156 at frequency ω=0, so that the sum of these group delays is 1.00000000, which is substantially equal to an integer number. The upsampling low-pass filter having the filter coefficients given in list (19) also satisfies the fourth condition as previously described. In this manner, the present embodiment employs a downsampling filter that satisfies the first through fourth conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. A description will be given of an eighth embodiment by referring to An upsampler A filter circuit The coefficients The filter circuit The coefficients In this manner, the present embodiment employs a downsampling filter that satisfies the first through fourth conditions previously defined, so that no pixel displacement occurs after 4:2:2-to-4:2:0 conversion and associated 4:2:0-to-4:2:2 conversion, and, also, chroma degradation does not worsen in cascade connections because of the attainment of the perfect reconstruction filter bank condition. Further, since the group delay of the 4:2:2-to-4:2:0 conversion is 0.25, compatibility with a conventional downsampling filter having preferable characteristics is attained in terms of phase characteristics. First Condition: the filter having these coefficients has such a group delay that a modulo-1 remainder of the group delay at frequency ω=0 is substantially equal to 0.25. Second Condition: there is at least one upsampling low-pass filter, such that the filter having the above-noted coefficients and the upsampling low-pass filter satisfy the perfect reconstruction filter bank condition within a predetermined error tolerance range.
If the answer obtained in step First Condition: there is at least one downsampling low-pass filter, such that the filter having the above-noted coefficients and the downsampling low-pass filter satisfy the perfect reconstruction filter bank condition within a predetermined error tolerance range.
If the answer obtained in step In the following, a description will be given of an example of top-field filter computation using rounded coefficients. Another example of filters A(z) and P(z) that simultaneously satisfy the first through fourth conditions is given in the following.
In a preferred method of computation, pixel values to be filtered are integer-multiplied by the integer part of the denominator of each filter coefficient as defined above for A(z) an P(z), and, then, the obtained products are added up, followed by adding 512 to the obtained sum for rounding purposes, and then discarding 10 bits (i.e., division by 1024). With this arrangement, proper filtering is achieved. All examples and conditional language recited herein are intended for pedagogical purposes to aid the reader in understanding the invention and the concepts contributed by the inventor to furthering the art, and are to be construed as being without limitation to such specifically recited examples and conditions, nor does the organization of such examples in the specification relate to a showing of the superiority and inferiority of the invention. Although the embodiment(s) of the present inventions have been described in detail, it should be understood that the various changes, substitutions, and alterations could be made hereto without departing from the spirit and scope of the invention. Patent Citations
Non-Patent Citations
Classifications
Legal Events
Rotate |